The reciprocal interplay between TNFα and the circadian clock impacts on cell proliferation and migration in Hodgkin lymphoma cells

Abstract

A bidirectional interaction between the circadian network and effector mechanisms of immunity brings on a proper working of both systems. In the present study, we used Hodgkin lymphoma (HL) as an experimental model for a type of cancer involving cells of the immune system. We identified this cancer type among haematological malignancies has having a strong differential expression of core-clock elements. Taking advantage of bioinformatics analyses and experimental procedures carried out in III- and IV-stage HL cells, and lymphoblastoid B cells, we explored this interplay and bear out diverse interacting partners of both systems. In particular, we assembled a wide-ranging network of clock-immune-related genes and pinpointed TNF as a crucial intermediary player. A robust circadian clock hallmarked III-stage lymphoma cells, differently from IV-stage HL cells, which do not harbour a properly functioning clockwork. TNF and circadian gene modulation impacted on clock genes expression and triggered phenotypic changes in lymphoma cells, suggesting a crucial involvement of core-clock elements and TNF in the physiopathological mechanisms hastening malignancy. Our results move forward our understanding of the putative role of the core-clock and TNF in the pathobiology of Hodgkin lymphoma, and highlight their influence in cellular proliferation and migration in lymphatic cancers.

Figure 1

The network of clock-immune related genes (NCIRG) connects elements of the circadian clock network and the immune system. An enrichment analysis on immune-related GO terms and KEEG pathways of the NCRG, and manual curation based on published data of immune-related genes with interactions with the CCN added a set (35) of immune-related genes (dark blue) to the network. Orange, light green and light purple elements represent the core-clock network (CCN, 20 elements), the extended core-clock network (ECCN) and the network of clock-related genes (NCRG), respectively. Grey lines represent the connection between genes previously published in the NCRG; green and red lines refer to positive and negative (respectively) interactions between newly added immune-related genes and the CCN, based on published data. Black arrows indicate the direction of each interaction. Dashed lines mark the interactions found in modified in vitro/in vivo models. It is important to notice that in some of the studies used for the construction of the network double KD or KO animals/cells were used. In such cases, it was not possible to specify the exact gene responsible for the reported effect on the immune genes, CRY, PER, ROR and REV-ERB were used to refer to the gene family. The direction of the effect is represented by an arrow for activation and a dead end arrow for the inhibition pointing to the affected gene.

Figure 2

Cancer cell line encyclopedia (CCLE) analysis revealed a circadian signature in Hodgkin lymphoma cell lines. (A) Represented is the number of genes, belonging to the “Circadian rhythms related genes” (WikiPathway), differentially expressed in the several haematopoietic and lymphoid tissue cancer cell lines present in the database. (B) Individual comparisons between the groups of cell lines regarding the number of differentially expressed genes of the pathway referred in A. (C,D) Heatmaps for relative expression levels of CCN genes in haematopoietic cancers and Hodgkin lymphoma cell lines, respectively. (E,G) LCL-HO, HD-MY-Z and L-1236 have differential circadian phenotypes. Shown are bioluminescence readouts for the promoter activity of Bmal1 (green) and Per2 (purple) over the course of 144 hours in LCL-HO (T = 29.33 ± 1.27 h; phase Bmal1 = 18.67 ± 0.3 h) (E) and the HL cell lines HD-MY-Z (T = 23.51 ± 0.06 h; phase Bmal1 = 11.64 ± 0.26 h) (F) and L-1236 (T = 32.61 ± 2.35 h; phase Bmal1 = 20.15 ± 0.3 h) (G). Periods and phases were calculated with ChronoStar software (n = 3, mean ± SEM).

Figure 3

The genome-wide landscape of oscillating genes in LCL-HO and HD-MY-Z revealed cell-specific circadian regulation in HL. (A) Depicted are expression values for core-clock genes for LCL-HO (blue) and HD-MY-Z (red) retrieved from the time course array data. (B) Oscillating genes present in LCL-HO and the HL cell line HD-MY-Z, respectively, phase ordered and normalized to values between −1 and 1. (C,D) Enrichment analysis (KEGG pathways) for the previously described genes in the different cell lines. (E) TNF relative gene expression retrieved from the array data in LCL-HO (p = 0.0006) and HD-MY-Z (p = 0.0104), respectively. (A,E) Based on the data from the time course HTA2.0 microarray experiments we performed an additional normalization in the range 1, −1. The normalized data was used to fit a curve with Local Polynomial Regression Fitting. The genes were analysed with the MetaCycle R package for a period set to 21 to 27 hours. The combined Metacycle p-values are displayed for each gene.

Figure 4

TNF affects the cell cycle and the expression of core-clock genes in HL. Gene expression, supernatant concentration of TNF and cell cycle analysis for the three cell lines (WT) after three days of stimulation with 50 ng/mL of recombinant human TNF, 0.1 uM dexamethasone and 25 uM of Forskolin. (A) Gene expression qPCR data (n = 3, mean ± SEM). (B) Supernatant concentration of TNF (n = 3, mean ± SEM). (C) Cell cycle analysis for the three cell lines (LCL-HO, the HL cell line HD-MY-Z and the HL cell line L-1236) (n = 3, mean ± SEM). (D) Gene expression data for Bmal1, Per2, Tnf, Cry1, Cry2 and Rev-erbα of LCL-HO, HD-MY-Z and L-1236 cell lines after shRNA KD (n = 3, mean ± SEM). The corresponding shRNA KD for the specific cell is represented under each plot. ^*p < 0.05, ^**p < 0.01, ^***p < 0.001.

Figure 5

Dysregulation of the core-clock and of Tnf led to differential cell cycle phenotypes in the HL cell line HD-MY-Z. (A) Gene expression qPCR data (n = 3, mean ± SEM). (B) TNF concentration in the supernatant (n = 3, mean ± SEM). (C) Cell cycle measurements after KD of Bmal1, Tnf, Per2 and Rev-erbα followed by stimulation for 3 days with 50 ng/mL of recombinant human TNF, 0.1 µM dexamethasone and 25 µM of forskolin (n = 3, mean ± SEM). (D) Proliferation analyses of HD-MY-Z cell line after shRNA KDs. (E,F) Apoptosis analysis of HD-MY-Z cell line after KDs in both non-stimulated and TNF (50 ng/ml) stimulated conditions, respectively (n = 3, mean ± SEM). Statistics for the comparison to non-stimulated (NS) control ^*p < 0.05, ^**p < 0.01, ^***p < 0.001.

Figure 6

Tnf and core-clock genes affected cell migration in the HL cell line HD-MY-Z. (A) Migration properties of pLKO.1 empty vector and KD HD-MY-Z cell lines (shBmal1, shTNF, shPer2 and shRev-erbα), measurements were obtained using a scratch wound assay for the IncuCyte S3 Live Cell System Analysis. Quantification was performed by measuring the relative wound confluence over the course of 34 hours. (n = 3, mean ± SEM). (B) Partial representation of the scratch wound assay for all HD-MY-Z KD derived cell lines in the course of 12–22 hours. Images were obtained with the IncuCyte S3 Software. Yellow mask indicates the wound border locations. Blue mask indicates the initial scratch wound area. (C) A model for the bi-directional role of TNF and the core-clock in regulating the molecular and cellular characteristics in HL. TNF regulates gene expression, cell cycle and apoptosis by interacting with the core-clock elements Bmal1, Per2 and Rev-erbα (orange circles). The grey area represents stimulatory effects of TNF on KD cell lines. TNF stimulation differentially affects the cell cycle phases of the KD cell lines, as illustrated by the different sizes and colour of each blue square (compared to pLKO.1). Cellular apoptosis is influenced by TNF stimulation in shRNA cell lines leading to its activation (shBmal, shRev-Erb) or inhibition (shPer2).